How the World's #1 VPN App Reached 1 Billion Downloads – Tanuj Chatterjee, Super Unlimited

Right after taking over the acquired app, read its App Store reviews and competitors' reviews, looking not for missing features but for complaints ('What are people complaining about?'). Fixed friction and service-quality problems first: picking the best server and protocol per user location, handling blocks (e.g. airlines blocking VPN traffic as bots). Treated the last 15% of quality as the moat since the first 85% of a VPN is a commodity.

No account creation; no heavy paywall that is hard to get around (paywall has a big X to close). First launch is a very quick setup; on later visits users tap Connect and leave for Netflix or their browser. Engagement is defined as intent to come back, not time spent in the app.

Rating prompt shown after value is delivered, not repeatedly or at the wrong moment; no further 'gymnastics' to farm ratings. Volume comes from multiplying a well-timed ask by a very large returning user base.

Ran methodical A/B tests of App Store screenshots: modernized designs, new colors, new content, plus per-region localized variants. Kept changes small because the app is #1 for 'VPN' and big listing changes are risky. Existing screenshots are simple and show streaming logos (HBO, Netflix).

Company had only a one-page site. In 2025, to prepare for the Windows launch, started programmatic content with Daydream, whose founder Tanuj knew and who had worked in VPN before. In-house content and SEO people continue it, but SEO is deliberately not a top channel because AI answers take search real estate.

Free users get unlimited VPN usage across dozens of countries and cities after watching an ad or two; nothing is throttled or taken away. Ads (interstitials, full-screen, native banners) are not optimized for profit; they are profitable in high-revenue countries (US, Western Europe) and loss-making elsewhere. Monetization kept restrained: 'leave 10% money on the table rather than bombard people'.

No ad right after install; the goal is for users to see the app's value first (and not hurt anyone who might convert). Paywall is soft with an easy close button.

Serve users in low-revenue or crisis countries (Turkey, Myanmar) even when ad match rates and CPMs are very low and the company loses money there. When bans or conflicts cause surges, add cloud capacity on top of the bare-metal fleet to absorb them.

Paid subscription adds: no ads, top-tier customer support, many more locations, access to 10G servers, and (since late 2025) one subscription covering all platforms (Windows, Mac, mobile, family devices). Plans to bundle in a second-phone-number app and eSIM.

Launched an eSIM app and simply showed an in-app ad for it to the large VPN user base. Next: target only likely travelers with eSIM-compatible phones, and run promotions before summer travel rather than bombarding everyone.

Only paid channel is Apple and Google search ads. Bids on specific keywords (beyond 'VPN') and specific markets, partly for brand defense; ROAS watched closely with a ceiling on spend. Uses extra push in countries where it isn't #1 yet (e.g. Android outside the US).

Head of support reports to head of product. Recurring questions and complaints become UX, communication, education, onboarding or service-quality fixes; regional ticket spikes (e.g. many service-quality tickets from Uganda) reach product immediately instead of via a monthly KPI review.

QA tests the long-lived VPN connection under real mobile conditions: hotel wifi to 3G/4G handoffs, elevators, switching cell towers on a bus. A QA engineer rode an elevator several times with a debug build to watch logs during disconnects.

Split App Store ratings by country; where a country rated notably lower than the US, investigated why and fixed the underlying issues (often service quality).

The per-country server/protocol selection algorithm would have taken months to build for all markets. Instead they shipped the version they believed right for five countries behind feature flags, planning to iterate on bugs or wrong assumptions and extend to other countries over time.

Start with standard cloud for speed to market. At scale, keep the control plane in mainstream cloud but run the data plane on bare-metal machines they control and wipe, since cloud bandwidth costs would 'lose our shirts'. Bare metal takes about a week to order, so cloud absorbs surges. Never cut costs in ways that hurt service quality.

Built and launched a Windows VPN (late 2025), held back until service quality was acceptable. Subscription covers all platforms. Main planned growth channel for it is product marketing inside the huge mobile user base rather than SEO.

Architected from the start so costs scale with paying end users; non-paying small developers use the product free.